CFX 12.1: y+ value for SST

ghorrocks · April 11, 2016, 01:47

My recommendation is to use SST unless you know a specific reason not to.

In your case the rotational effects might cause a problem, but I would try with the curvature correction model with SST before I considered any other turbulence model.

ngoc_tran_bao · April 15, 2016, 04:01

Thank you so much for your recommend, Ghorrocks. I have tried to refine the mesh at all walls by creating 10 layers of inflation to get the value Y+ of 1 and apply SST turbulence model. It is harder to get convergence than K-e model but I believe it will be alright. By the way, I have another question about head loss, you know beside friction there are many types of head loss in the pump such as volumetric loss, hydraulic loss and so on. Is there any way to apply those loss in our simulation i order to make the result to be more accurate, Sir?

ghorrocks · April 15, 2016, 07:01

The whole idea of CFD is that you do not need to define these factors as the CFD model works it out based on the flow conditions.

If you wish to apply all those loss factors then you are doing a system level model, not CFD.

ozturk · April 28, 2016, 15:23

Hello,
I am analizing 3d naca airfoil. I use k-w SST and my Re 4.1e+7. I want to learn drag and lift over airfoil. Airflow velocity is 210 m/s. I desired y+ value 1 and calculate my first wall (infilation) height is 5.2e-6 . My infilation number is 45. After I runned my case, I plotted my y+ value over airfol is between 0,05 and 0,68. Is this value suitable for calculating right drag and lift?
Thank you.
Regards

ghorrocks · April 28, 2016, 20:51

You cannot say from a simple y+ number whether your mesh is adequate. You should do a mesh sensitivity study to determine whether your mesh is accurate enough for your purposes or not.

ozturk · April 29, 2016, 03:56

Thanks for your quick reply,
You said I should do a mesh sensitivity study; I should run different mesh elements number.
I calculated my case between 2000000 and 4500000 elements and then my max skewness nearly 0.95 . Before I said you, I desired my y+ value 1; but my results between 0.05 and 0.65, it is under 1 . It is not exactly 1. I want to ask these:
1. My solution model k-w STT is true?
2. My desired y+ value 1 is true?
3. After calculation I have to catch 1 y+ value, if it is lower 1, is it problem?
Thank you.
Regards.

ghorrocks · April 29, 2016, 04:37

y+<1 is a guide for what is required for SST to do integration to the wall with reasonable accuracy. But note that if you can use wall functions (and many flows can) then you are aiming for y+>11. So do not simply assume you need a small y+ and wall functions are much easier to use and work for many flows.

In my experience as the mesh is refined and the y+ gets smaller the results get more accurate. But there will be a point where numerical round off errors start becoming important and that means that further mesh refinement will lead to divergence. So there is a lower limit on mesh refinement. And the lower limit also varies between simulations.

ozturk · April 29, 2016, 05:07

I understood that: I think my way is true. Because I do not use k-epsilon ( wall function ). I use k-w STT so for my y+ value 1 is right desired value.
Thank you again.
Regards

ghorrocks · April 29, 2016, 08:55

Because you are using SST does NOT mean you then have to have y+=1. If you are using automatic wall functions it will automatically transition from wall functions to integrating to the wall. You should try a y+ which will give wall functions and see if it works in your situation.

ozturk · June 6, 2016, 08:55

Hi,
Again I am here

Maybe you can remember my model is 3d airfoil. I analize cd nad cl coefficient. I want to ask that I write the area which looking airfoil through y axis. I know this is true for cl coefficient but I am not sure for cd.
For cd it maybe looking through x axis.
Thank you.

Blauwal · August 21, 2019, 04:36

Hi everyone!

I'm barely new in CFD and currently working with cfx on my first study. Because i have to write a report for my university I would like to explain why I took a y + value of 1.

I tried to understand the theory of delta y the viscose sublayer etc. but in short words or better related on any paper is there anything you could recommend? i do not want to explain it in detail but relate me on something.

Thank you in advance!!!

karachun · August 21, 2019, 06:23

https://www.youtube.com/watch?v=fJDYtEGMgzs

April 28, 2016, 15:23	Is it right value?	#24
ozturk New Member ozturk sancaktar Join Date: Jun 2015 Posts: 6 Rep Power: 10	Hello, I am analizing 3d naca airfoil. I use k-w SST and my Re 4.1e+7. I want to learn drag and lift over airfoil. Airflow velocity is 210 m/s. I desired y+ value 1 and calculate my first wall (infilation) height is 5.2e-6 . My infilation number is 45. After I runned my case, I plotted my y+ value over airfol is between 0,05 and 0,68. Is this value suitable for calculating right drag and lift? Thank you. Regards

April 29, 2016, 04:37		#27
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,727 Rep Power: 143	y+<1 is a guide for what is required for SST to do integration to the wall with reasonable accuracy. But note that if you can use wall functions (and many flows can) then you are aiming for y+>11. So do not simply assume you need a small y+ and wall functions are much easier to use and work for many flows. In my experience as the mesh is refined and the y+ gets smaller the results get more accurate. But there will be a point where numerical round off errors start becoming important and that means that further mesh refinement will lead to divergence. So there is a lower limit on mesh refinement. And the lower limit also varies between simulations. ozturk likes this.

April 29, 2016, 08:55		#29
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,727 Rep Power: 143	Because you are using SST does NOT mean you then have to have y+=1. If you are using automatic wall functions it will automatically transition from wall functions to integrating to the wall. You should try a y+ which will give wall functions and see if it works in your situation. ozturk likes this.

August 21, 2019, 04:36	Why should y+ be around 1?	#31
Blauwal New Member Sven Join Date: Mar 2019 Posts: 1 Rep Power: 0	Hi everyone! I'm barely new in CFD and currently working with cfx on my first study. Because i have to write a report for my university I would like to explain why I took a y + value of 1. I tried to understand the theory of delta y the viscose sublayer etc. but in short words or better related on any paper is there anything you could recommend? i do not want to explain it in detail but relate me on something. Thank you in advance!!!

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Pros and Cons for CFX, CFdesign, COMSOL	Val	Main CFD Forum	3	June 10, 2011 02:20
Importing solutions in CFX.	Alphonso	CFX	1	August 1, 2008 14:01
PhD using CFX	Rui	CFX	9	May 28, 2007 05:59
CFX 10 VS CFX 11 for combustion	Jonathan Lemay	CFX	2	May 9, 2007 11:58
CFX 4.4 installation problem	Pandu Sattvika	CFX	1	December 1, 2001 04:07

April 11, 2016, 01:47		#21
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,727 Rep Power: 143	My recommendation is to use SST unless you know a specific reason not to. In your case the rotational effects might cause a problem, but I would try with the curvature correction model with SST before I considered any other turbulence model.

April 15, 2016, 04:01		#22
ngoc_tran_bao Member ngoc tran bao Join Date: Jan 2016 Posts: 35 Rep Power: 10	Thank you so much for your recommend, Ghorrocks. I have tried to refine the mesh at all walls by creating 10 layers of inflation to get the value Y+ of 1 and apply SST turbulence model. It is harder to get convergence than K-e model but I believe it will be alright. By the way, I have another question about head loss, you know beside friction there are many types of head loss in the pump such as volumetric loss, hydraulic loss and so on. Is there any way to apply those loss in our simulation i order to make the result to be more accurate, Sir?

April 15, 2016, 07:01		#23
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,727 Rep Power: 143	The whole idea of CFD is that you do not need to define these factors as the CFD model works it out based on the flow conditions. If you wish to apply all those loss factors then you are doing a system level model, not CFD.

April 28, 2016, 20:51		#25
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,727 Rep Power: 143	You cannot say from a simple y+ number whether your mesh is adequate. You should do a mesh sensitivity study to determine whether your mesh is accurate enough for your purposes or not.

April 29, 2016, 03:56		#26
ozturk New Member ozturk sancaktar Join Date: Jun 2015 Posts: 6 Rep Power: 10	Thanks for your quick reply, You said I should do a mesh sensitivity study; I should run different mesh elements number. I calculated my case between 2000000 and 4500000 elements and then my max skewness nearly 0.95 . Before I said you, I desired my y+ value 1; but my results between 0.05 and 0.65, it is under 1 . It is not exactly 1. I want to ask these: 1. My solution model k-w STT is true? 2. My desired y+ value 1 is true? 3. After calculation I have to catch 1 y+ value, if it is lower 1, is it problem? Thank you. Regards.

April 29, 2016, 05:07		#28
ozturk New Member ozturk sancaktar Join Date: Jun 2015 Posts: 6 Rep Power: 10	I understood that: I think my way is true. Because I do not use k-epsilon ( wall function ). I use k-w STT so for my y+ value 1 is right desired value. Thank you again. Regards

June 6, 2016, 08:55		#30
ozturk New Member ozturk sancaktar Join Date: Jun 2015 Posts: 6 Rep Power: 10	Hi, Again I am here Maybe you can remember my model is 3d airfoil. I analize cd nad cl coefficient. I want to ask that I write the area which looking airfoil through y axis. I know this is true for cl coefficient but I am not sure for cd. For cd it maybe looking through x axis. Thank you.

August 21, 2019, 06:23		#32
karachun Senior Member Join Date: Nov 2015 Posts: 244 Rep Power: 11	https://www.youtube.com/watch?v=fJDYtEGMgzs